LA Wildfires Raise Burning Questions About AI’s Data Center Water DrainLA Wildfires Raise Burning Questions About AI’s Data Center Water Drain

Apocalyptic scenes from greater Los Angeles, California’s continuing wildfire devastation raise serious questions about IT’s growing need for a resource in short supply: water. The explosion of power-hungry AI models is a growing strain on water resources, even as the industry makes strides in mitigation efforts.

Many factors -- from water shortages due to an ongoing drought, to infrastructure restraints -- led to a shortage of water and water pressure in fire hydrants throughout Los Angeles County. The shortages fueled partisan finger-pointing over blame. Water is increasingly becoming a major stress point for governments as IT needs increase.

Three Democratic California lawmakers introduced four separate bills last week aimed at slowing AI and data center water consumption. One of the bills’ authors, Assemblymember Dian Papan, told Politico, “Water’s a limited resource. I’m trying to make it so we are prepared and ahead of the curve as we pursue new technology.”

Providing a snapshot of increasing data center water use needs, the US Department of Energy’s report on the country’s data center energy use pegs total 2023 water use at 66 billion liters, up from 21.2 billion liters in 2014. And that’s just for direct consumption to cool data centers themselves -- water needed to cool power plants supplying electricity to data centers, also adds to the total.

Related:How Operating Models Need to Evolve in 2025

But transparency on water use is an issue. About 50% of organizations do not collect water usage data for data center operations, according to Statista.

Data Center Map counts 286 data centers in California, including 69 in Los Angeles.

AI Proliferation Driving Increased Water Needs

Artificial intelligence drove about 20% of new data center demand over the last year, according to a report from global commercial real estate firm JLL. The market for colocation data centers, which soak up some of the highest water use rates, doubled in size over the past 4 years, according to the report. Data creation is expected to increase at a compound annual growth rate of 32% through 2030.

The arms race to develop AI tools in the enterprise has driven excitement and fear about the emerging technology.

Much of the hype around AI ethics has revolved around potential existential threats. Energy consumption and water use may not be a topic quite as scintillating as impending robot doomsday scenarios, but experts say the environmental impacts may pose the most immediate threat.

“It’s heartbreaking to witness the aftermath of the LA fires and how they’ve exposed critical water infrastructure challenges,” says Manoj Saxena, CEO and founder of Responsible AI Institute and information Insight Circle member. “While we often debate the existential threats of AI, the immediate reality is its growing environmental impact -- particularly on carbon emissions and water consumption.”

Related:Why Your Business May Want to Shift to an Industry Cloud Platform

Pointing to statistics from the World Economic Forum, Saxena says global AI demand could push water usage to an “astonishing” 1.7 trillion gallons of annual water use. “The fact that 20% of these servers already rely on water from stressed watersheds is a wake-up call.”

Water Saving Strategies: Can We Keep Up?

There are many water saving techniques data centers are deploying, including immersion cooling (submerging servers in liquid), free cooling (using outside air in colder climates), direct-to-chip cooling, and more.

But as more sustainable techniques come online, the need for much more powerful data center servers could cancel out those efforts. Older data centers cannot keep up with the computing needs of new AI systems. And upgrades mean more strain on water resources, so experts are pushing for initiatives to keep up with increasing demand.

Companies are racing to adopt more sustainable data center plans. Microsoft, for example, is moving forward with new data center designs that use chip-level cooling to consume no water. “This design will avoid the need for more than 125 million liters of water per year per datacenter,” Steve Solomon, Microsoft’s vice president for datacenter infrastructure engineering, said in a blog post.

Related:Ensure Your Organization’s Cloud Is Ready for AI Innovation

Considering Microsoft reported its cloud data centers had soaked up 6.4 million cubic meters of water in 2022 (a 34% increase from the year prior), canceling out water use would be a big win. But overall, tech companies have struggled to meet previously set sustainability goals as generative AI unexpectedly took off with the release of ChatGPT.

But RAI’s Saxena says more needs to be done -- and quickly.

“We need to act now to ensure AI’s growth doesn’t come at the cost of our planet,” Saxena says. “This means adopting water-efficient cooling technologies, capping water use in drought-prone regions, promoting closed-loop cooling systems, incentivizing renewable-powered AI operations, and fostering public-private partnerships to set sustainable infrastructure policies.”